Seite - 228 - in Adaptive and Intelligent Temperature Control of Microwave Heating Systems with Multiple Sources

A. Appendix N1,1 N1,2 N2,1 N2,2 N2,3 N3,1 Input Layer Hidden Layer Output Layer W11,1 YNN,1 u1(k) u2(k) W13,4 Bias = 1 N3,2 YNN,2 W22,4 Bias = 1 W22,1 FigureB.1. Errorflowintheoutput layer. A.2. DerivationofBackpropagation Algorithm Case1: Forweightsconnectingto theoutput layer If thedestinationofthelinkis intheoutput layer,suchasthebluelink in figure B.1, any error of the weight only affects one output and the corresponding weight update is simple and straightforward. For the destination node j in the output layer, according to the definition in table 3.3 its inputcanberepresentedby zoj(k) = NL∑ i=1 woj,i(k)x L i (k), (B.19) 228